Generative AI in Everyday Academic Work: An Exploratory Survey of Usage Patterns, Perceived Effectiveness, and Concerns among Indonesian University Students

Abstract

Background. Generative artificial intelligence (generative AI) tools have diffused rapidly into university study, yet, in the literature reviewed for this study, most empirical evidence concerns designed interventions or Western student populations. How students in Southeast Asia draw on these tools in unsupervised, self-directed academic work is sparsely documented in that literature. Objective. This exploratory study describes self-reported generative AI use, perceived effectiveness, and the benefits and concerns articulated by university students, primarily in Indonesia. Methods. A cross-sectional online survey was administered via Google Forms in June 2026. Twenty-four students accessed the survey; one declined consent, yielding 23 analysable cases (21 Indonesian, 2 Chinese). The instrument comprised seven usage-frequency items referring generically to “AI tools” and seven perceived-effectiveness items, both on five-point Likert scales (? = .92 across the 14 items), and two open-ended questions. Analysis was restricted to descriptive statistics, one descriptive correlation coefficient (not interpreted inferentially), and reflexive thematic analysis. No inferential group comparison was attempted. Results. Students reported using generative AI most often to understand difficult course materials (M = 3.65, SD = 1.27) and to brainstorm ideas or generate explanations (M = 3.52, SD = 1.16), and least often as part of a daily study routine (M = 2.65, SD = 1.15). Perceived effectiveness was highest for comprehension (M = 3.61, SD = 1.34) and lowest for overall learning achievement (M = 3.00, SD = 1.13). The two composite scores correlated at r = .87 in this sample; because both derive from a single instrument at one occasion, the coefficient is reported descriptively and is not interpreted as evidence of a relationship between distinct constructs. Thematic analysis of the open responses identified time efficiency (27% of 22 respondents) as the most frequently reported benefit, with personalized tutoring, brainstorming support, and improved comprehension each cited by 18% of 22 respondents. The most frequently reported concern was over-reliance diminishing independent thinking (37% of 19 respondents), followed by doubts about the accuracy of AI output (26% of 19 respondents) and difficulty formulating effective prompts (16% of 19 respondents). Conclusions. Within the limits of a very small convenience sample, the findings describe generative AI used as a task-specific aid concentrated at points of comprehension difficulty rather than as a constant study companion, alongside widely voiced apprehension about dependence. The results are exploratory and descriptive, intended to generate rather than test hypotheses; if replicated in larger samples, they would suggest that AI literacy provision, instruction in prompting, and assessment designs that keep students cognitively engaged merit examination, and they indicate directions for adequately powered confirmatory research.

Introduction

The study explores how university students use generative artificial intelligence (AI) for academic purposes, their perceptions of its effectiveness, and the benefits and concerns they associate with it. Although AI tools such as ChatGPT, DeepSeek, and Gemini can support learning, their educational impact depends on how students use them. Critical engagement with AI-generated information may improve understanding and writing, while excessive reliance may weaken independent thinking and deep learning.

The study addresses a research gap concerning self-directed AI use among university students, particularly in Indonesia and Southeast Asia, where systematic survey evidence remains limited. It employed an exploratory, cross-sectional survey of 23 university students, including 21 from Indonesia and two from China. Data were collected through Likert-scale questions and open-ended responses, covering AI usage frequency, perceived effectiveness, benefits, and challenges.

The findings indicate that students most frequently use AI to understand difficult course materials, brainstorm ideas, and generate explanations. Comprehension support received the highest usage rating (M = 3.65), whereas incorporating AI into daily study routines received the lowest (M = 2.65). The overall mean usage score was 3.22 out of 5, suggesting moderate engagement with AI tools.

Regarding perceived effectiveness, students rated improved understanding of learning materials highest (M = 3.61). Perceived improvements in academic performance, confidence, efficiency, and assignment quality received moderate ratings. Overall learning achievement received the lowest effectiveness rating (M = 3.00), indicating that students were more confident about AI helping them understand content than about it improving their broader academic outcomes.

The main benefits identified were time efficiency and increased productivity, followed by personalized explanations, brainstorming, and improved comprehension. Students also reported that AI could increase confidence and make unfamiliar topics easier to approach.

However, the most frequently reported concern was over-reliance on AI and the potential decline of independent thinking and critical reasoning. Other concerns included the accuracy and reliability of AI-generated information, difficulties writing effective prompts, limited access or usage restrictions, and occasional issues involving academic integrity, privacy, motivation, and the limitations of AI systems.

The study also found a strong descriptive correlation between reported AI usage and perceived effectiveness (r = .87). However, because both measures came from the same self-report questionnaire, this result cannot establish a causal relationship or confirm that the two measures represent distinct constructs.

Conclusion

This exploratory descriptive survey of 23 university students, 21 of them in Indonesia, offers an early description of how generative AI enters self-directed academic work within a small convenience sample. Within this small sample, AI functioned as a task-specific aid concentrated where new material must be understood and where ideas must be generated, rather than as a constant companion to study. Students endorsed its contribution to comprehension more readily than any broader claim about their attainment, and the concern they raised most often, in their own words, was that the convenience they valued might be displacing their capacity to think independently. If replicated in larger samples, these findings would suggest that responses to that tension might centre on instructional design that keeps students in an active relation to the material: assessment that requires interrogation of AI output rather than its reproduction, explicit provision in evaluating output and constructing prompts, and institutional guidance concrete enough to distinguish support from substitution. Whether such approaches are effective was not examined here. The prominence of prompt formulation as a self-identified difficulty is a finding that warrants further investigation. These conclusions are provisional in proportion to the evidence behind them. Twenty-three respondents recruited by convenience cannot establish how Indonesian students in general use AI. What the study can offer is a transparently reported starting point and a set of specific questions for adequately powered work: whether the concentration of use at comprehension replicates in larger samples; whether awareness of over-reliance predicts any protective behaviour; and whether prompt formulation constitutes a distinct and teachable component of AI literacy. Longitudinal designs, behavioural trace data, and instruments capable of distinguishing usage from appraisal will be needed to answer them.

References

[1] Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3(2), 77–101. https://doi.org/10.1191/1478088706qp063oa [2] Luo, J., Zheng, C., Yin, J., & Teo, H. H. (2025). Design and assessment of AI-based learning tools in higher education: A systematic review. International Journal of Educational Technology in Higher Education, 22(1), 1–27. https://doi.org/10.1186/s41239-025-00540-2 [3] Madanchian, M., & Taherdoost, H. (2025). Examining critical factors in selecting AI tools for educational success. Procedia Computer Science, 263, 923–933. https://doi.org/10.1016/j.procs.2025.07.111 [4] Mallik, S., & Gangopadhyay, A. (2023). Proactive and reactive engagement of artificial intelligence methods for education: A review. Frontiers in Artificial Intelligence, 6, 1151391. https://doi.org/10.3389/frai.2023.1151391 [5] Memari, M., & Ruggles, K. (2025) (preprint). Artificial intelligence in elementary STEM education: A systematic review of current applications and future challenges. arXiv. https://doi.org/10.48550/arXiv.2511.00105 [6] Teng, M. F. (2025a). Understanding EFL student writers\' metacognitive awareness in utilizing ChatGPT. System, 135, 103848. https://doi.org/10.1016/j.system.2025.103848 [7] Teng, M. F. (2025b). Metacognitive awareness and EFL learners\' perceptions and experiences in utilising ChatGPT for writing feedback. European Journal of Education, 60(2), e12811. https://doi.org/10.1111/ejed.12811 [8] Wang, S., Wang, F., Zhu, Z., Wang, J., Tran, T., & Du, Z. (2024). Artificial intelligence in education: A systematic literature review. Expert Systems with Applications, 252, 124167. https://doi.org/10.1016/j.eswa.2024.124167 [9] Younas, M., El-Dakhs, D. A. S., & Noor, U. (2025). The impact of artificial intelligence-based learning tools in academic innovation: A review of DeepSeek, GPT, and Gemini (2020–2025). Frontiers in Education, 10, 1689205. https://doi.org/10.3389/feduc.2025.1689205

Copyright

Copyright © 2026 Widhia Andhini Nur Athirha , Natasya Amalia Nur Azizah, Calvanny Lambue Rezananda. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.